---
title: "What the past.dev MCP server is"
description: "A single secure address that puts your own past.dev account in front of an MCP assistant, so you can work on your integration from the assistant you already work in."
canonical: https://past.dev/docs/mcp/overview
last-updated: 2026-08-26
---
# What the past.dev MCP server is

> A single secure address that puts your own past.dev account in front of an MCP assistant, so you can work on your integration from the assistant you already work in.

Product: past.dev MCP Server. Source: https://past.dev/docs/mcp/overview

MCP, the Model Context Protocol, is an open standard for connecting AI assistants to real systems. past.dev runs an MCP *server*. You point your assistant at it once, sign in with your normal past.dev account, and from then on the assistant can reach that account directly.

> **Read this before you plan an integration around it**
>
> This server is for **you**, the developer with a past.dev account. It is not a memory feed for your product's end users, and pointing your users at it is not a supported design: past.dev does not hold your user directory and cannot tell your users apart.
>
> To give your own users memory inside their assistant, you run your own MCP server over the Memory API. It is a small amount of code, you keep control of who your users are, and section 04 walks through it.

What it is good for, working on an integration:

- “What has past.dev learned about me so far? I want to see what a real record looks like before I write the mapping.”
- “Save this as a memory on my account so I can check what `/recall` returns for it.”
- “I am about to ingest a batch. Show me what is already there so I do not duplicate it.”

Everything runs as **you**: the same account, the same project, the same permissions you have in the past.dev console. If you cannot see something in past.dev, the assistant cannot see it either.

### What connects to it

Anything that supports remote MCP servers over Streamable HTTP with OAuth. That includes Claude (desktop, web and Claude Code), Cursor, VS Code with GitHub Copilot, Windsurf, Zed, and a growing list of others. Section 2 has the exact steps for each.

### The documentation server is separate

There is a second, read-only MCP server at `past.dev/mcp` that serves this documentation and needs no sign-in at all. Point an assistant at it while you are writing code against the API. It carries no account data, so the two are never confused.

> **Good to know**
>
> past.dev is **not yet listed in Claude's connector directory**, so you add it manually as a custom connector. It takes about thirty seconds and the result is identical to a directory listing: same OAuth flow, same security. Nothing is missing or degraded.